JPH0743309A - Pattern inspection method - Google Patents

Abstract

PURPOSE:To perform pattern inspection without affected by fluctuation in manufacture by, within allowance in manufacture, inspecting similarity among pixels for each pixel in comparison pixel, and selecting, for comparison, pixels to be compared with comparison image, one by one. CONSTITUTION:Common alignment (5) between pixels is performed so that an inspection image 4 and a comparison image 3 are adapted to an image. Between the inspection image 4 and comparison image 3, the difference of c pixel and d pixel in line thickness when compared with the comparison image 3 for a line element (a) and a line element (b) that correspond to each other. However, even if this is within allowance in manufacture, difference appears when comparison is made (7) after alignment of the inspection image 4 with the comparison image 3. When this is inspected by pattern inspection method, comparison difference disappears almost entirely if c and d pixels are within evaluation range + or -L of similarity. Therefore, by setting evaluation range of similarity to within allowance, inspection (8) is done without affected by fluctuation in manufacture.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection method, and more particularly to a pattern inspection method for detecting a fine pattern such as an LSI wafer or TFT with high resolution and comparing the patterns to inspect for defects.

[0002]

2. Description of the Related Art As a conventional technique, for example, Japanese Patent Laid-Open No.
There is a pattern recognition method disclosed in Japanese Patent Laid-Open No. 194702. The pattern recognition method described above includes a pixel unit matching unit that aligns a pattern to be inspected and the pattern and a previously stored pattern or a separate detection pattern between screens on a pixel unit basis, and a position using the pixel unit matching unit. The difference between the sub-pixel matching unit that re-aligns each aligned pixel in the aligned pattern with an accuracy of a pixel or less and the pattern that is aligned with the accuracy of a pixel or less in the sub-pixel matching unit is obtained. It is configured to include a difference image extraction unit and a defect determination unit that determines a defect from the value of the difference image extraction unit.

Next, a conventional pattern inspection method will be described with reference to the drawings.

FIG. 3 is a block diagram of a pattern inspection apparatus using a conventional pattern inspection method. The pattern inspection apparatus shown in FIG. 3 records a detection unit 31 including illumination, a positioning mechanism, and a photoelectric converter, an A / D 33 that digitally converts a video signal from the detection unit 32, and a digital signal of the A / D 33. Image memory unit 34, and image memory unit 34
A comparison image extracting unit 35 for extracting an image to be compared from
And a pixel unit matching unit 36 that performs pixel unit positioning between the A / D 33 and the inspection image of the comparison image extracting unit 35.
And a sub-pixel matching unit 37 that performs matching with matching accuracy of pixels or less between matching pixels of the pixel-unit matching unit 36, and a sub-pixel alignment unit 33 that complements accuracy of matching pixels or less with matching information from the sub-pixel matching unit 37. And a difference image extraction unit 39 that obtains a difference with respect to the alignment result of the sub-pixel alignment unit 33.
And a defect determination unit 40 that determines a defect in the difference image of the difference image extraction unit 39, and an overall control unit 32 that controls the entire image.

In the pattern inspection, the inspection image f (x, y) obtained by digitally converting the data from the detection unit 31 by the A / D 33 and the corresponding image from the image memory 34 are compared image extraction unit 35. With respect to the comparison image g (x, y) obtained by extracting in step (1), the pixel unit matching unit 36 first performs alignment between the inspection image f (x, y) and the comparison image g (x, y), for example, f ( x (y,) for g (x,
If there is a positional deviation of dx in the x direction and dy in the y direction between the images of y), g (x, y) is calculated in the pixel unit matching unit.
Is corrected as g (x + dx, y + dy). Next, the sub-pixel matching unit 37 and the sub-pixel alignment unit 38 perform the sub-pixel alignment for each pixel by the least square method. The difference between the images as the result of the alignment is determined by the difference image extraction unit 39, and the defect determination unit 40 determines the defect based on a threshold value specified in advance.

[0006]

According to the conventional pattern inspection method, when the pattern width changes by several pixels of the detection resolution due to fluctuations in the manufacturing process or the like, a defect occurs even if it is a manufacturing allowable value. There was a problem.

[0007]

A pattern inspection apparatus of the present invention comprises an inspection image memory in which an inspection target pattern is recorded, a pre-recorded comparison image memory for comparison inspection with the pattern in the inspection image memory, and An inter-pixel similarity determiner for determining the similarity of the inspection image memory with respect to each pixel of the inspection image memory and the comparison image memory of the comparison image memory, and a determination result from the similarity determiner A pixel shift selector that selects a pixel of the inspection image memory having a high degree of similarity, and a difference between the pixel of the inspection image memory and the pixel of the comparison image memory selected by the pixel shift selector. An inter-pixel difference device and a defect determination device for determining a defect based on the calculation result of the inter-pixel difference device are provided.

[0008]

The present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention. The pattern inspection apparatus shown in FIG. 1 includes a detection unit 1 including illumination, a positioning mechanism, and a photoelectric converter, an A / D 2 that digitally converts a video signal from the detection unit 1, and an A / D
The image information from D2 is stored in the inspection image memory 4 and the comparison image memory 3 which are recorded by the position information (here, the comparison image memory 3 may also be recorded with a pattern previously recorded and placed. .) It is determined by the inter-pixel similarity determiner 5 and the inter-pixel similarity determiner 5 that determine the closest pixel in the inspection image memory 4 and the comparison image memory 3 from the inspection image memory 4. An inter-pixel differencer 7 for selecting the inspection pixels of the inspection image memory 4 by the pixel shift selector 6 according to the amount and comparing the inspection pixels of the comparison image memory 3;
It is configured by a defect determiner 8 that determines a defect with a threshold value set in advance for the comparison result of the inter-pixel difference unit 7.

In the above structure, the pattern comparison is performed at the position (X, Y) for each pixel of the inspection image f (x, y) and the comparison image g (x, y). The difference is calculated by the differentiator 7 as (X, Y) = abs [g (X, Y) -f (X + χ, Y + ψ)] (1). The shift amount (χ, ψ) is calculated by the inter-pixel similarity determiner 5 using the similarity evaluation function H.

[0011]

As a method of evaluating the degree of similarity, the pixel having the closest difference in gray value of the image of the designated area with the determination target pixel as the center is selected, or the cross-correlation of the vertical and horizontal waveforms with the determination target pixel as the center is most selected. A method such as selecting a high pixel can be considered.

For example, according to the present invention, when the inspection image of FIG. 2 (a) and the comparison image of FIG. 2 (b) are aligned between general pixels adapted to the image, FIG. 2 (c) is obtained. In the inspection image and the comparison image, the line widths of the corresponding line elements a and b are different from each other in the line widths of −c pixels and d pixels with reference to the comparison image.
However, even if this is within the manufacturing allowable range, when the inspection image FIG. 2 (a) and the comparison image FIG. 2 (b) are compared after alignment, the difference of FIG. 2 (d) occurs. When this is inspected by the pattern inspection method of the present invention, the similarity evaluation range ± L
If there are c and d pixels in the range, the result of FIG. 2E is obtained, and if the similarity evaluation range is set to the manufacturing allowable range, the inspection can be performed without being affected by the manufacturing fluctuation.

[0015]

As described above, according to the present invention, the similarity between pixels is inspected for each comparison pixel within the manufacturing allowable range, the comparison image is used to sequentially select the comparison pixels, and the comparison is performed. In addition, the pattern inspection can be performed without being affected by manufacturing fluctuations.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

2A to 2E are pattern waveform diagrams.

FIG. 3 is a block diagram showing a conventional example.

[Explanation of symbols]

 1,31 Detection unit 2,33 A / D 3 Comparison image memory 4 Inspection image memory 5 Pixel similarity determiner 6 Pixel shift selector 7 Pixel differencer 8,40 Defect determiner 32 Overall control unit 34 image Memory unit 35 Comparative image extractor 36 Pixel unit matching unit 37 Sub pixel matching unit 38 Sub pixel alignment unit 39 Difference image extracting unit a, b Line elements c, d Line width difference pixels

Claims (4)

[Claims] 1. An inspection image memory in which a pattern to be inspected is recorded, a comparison image memory previously recorded for comparison inspection with a pattern in the inspection image memory, the inspection image memory and the comparison image memory. An inter-pixel similarity determiner for determining the similarity of the inspection image memory with respect to each pixel of the comparison image memory, and a pixel of the inspection image memory having a high degree of similarity based on the determination result from the similarity determiner Of the pixel shift selector for selecting the pixel shift selector, the pixel difference selector for calculating the difference between the pixel of the inspection image memory and the pixel of the comparison image memory selected by the pixel shift selector, A pattern inspection apparatus comprising: a defect determiner for determining a defect based on a calculation result. 2. (A) A procedure of recording a comparison image for comparison inspection with a pattern to be inspected in a comparison image memory in advance, (B) A procedure of recording the pattern to be inspected in the image memory for inspection, C) A pixel-to-pixel similarity determination procedure in which a pixel having the closest similarity between the comparison image memory and the inspection image memory is determined from the inspection image memory, and the determination amount is obtained, (D) Based on the determination amount A pixel shift selection procedure of selecting an inspection pixel of the inspection image memory, (E) comparing the selected inspection pixel of the inspection image memory with a pixel of the comparison image memory, and A pattern inspection method comprising: a procedure for detecting a difference; and (F) a defect determination procedure for determining a defect based on a preset threshold value with respect to the difference. 3. The pattern inspection method according to claim 2, wherein, as the inter-pixel similarity determination procedure, a pixel having a gray scale value closest to that of an image of a designated area is selected around the determination target pixel. 4. The pattern inspection method according to claim 2, wherein, as the inter-pixel similarity determination procedure, a pixel having the highest cross-correlation of vertical and horizontal waveforms centering on the determination target pixel is selected.